gLite 3.1
lcg-CE - Update to version 3.1.11-0
|
| Date |
22.04.08 |
| Priority |
High |
Description
The update of the lcg-CE service includes changes in Globus jobmanager and
GASS cache.
These modifications improve the performance of the lcg-CE by a factor of two to three.
The update modifies the behavior of the Globus jobmanager machinery, optimizing it in
two ways: The first one is the limitation of the number of concurrently running jobmanager queries
decreasing load on the filesystem and the batch system and thus allowing queries to run faster.
The second modification is the conversion of certain Globus Perl scripts to memory-persistent
daemons eliminating the necessity to compile heavy Perl code on every invocation. The Globus
jobmanagers communicate with Perl daemons over UNIX domain sockets using a tiny compiled client
written in C. These clients replace original Globus Perl scripts, so the interface change is
transparent for both Globus and users.
Two new daemons are introduced by this update. They are globus-gass-cache-marshal and
globus-job-manager-marshal. The relevant configuration files can be found under
/opt/globus/etc. The log file location is /opt/globus/var/log.
The daemons should be started using init scripts before the globus-gatekeeper on boot and
they should be stopped after the globus-gatekeeper on shutdown. They intercept HUP signal,
so it is possible to rotate log files and change running configuration without restart.
Please also have a look at the list of known issues.
Updated rpms
The RPMs can be updated using yum via
Service reconfiguration after update
Service must be reconfigured.
Service restart after update
Not needed.
How to apply the fix
- Update the RPMs (see above)
- Update configuration (see above)
- Restart the service if necessary (see above)
|
|
